光谱变换和光谱分辨率对土壤有机质含量估测精度的影响  

Effects of spectral transformation and spectral resolution on the estimation accuracy of soil organic matter content

作　　者：张霄羽 姚艳敏[1] 颜祥照 ZHANG Xiao-yu;YAO Yan-min;YAN Xiang-zhao(State Key Laboratory of Efficient Utilization of Arid and Semi-arid Arable Land in Northern China/Institute of Agricultural Resources and Regional Planning,Chinese Academy of Agricultural Sciences/Key Laboratory of Agricultural Remote Sensing,Ministry of Agriculture and Rural Affairs,Beijing 100081)

机构地区：[1]北方干旱半干旱耕地高效利用全国重点实验室/中国农业科学院农业资源与农业区划研究所/农业农村部农业遥感重点实验室,北京100081

出　　处：《中国土壤与肥料》2023年第3期184-193,共10页Soil and Fertilizer Sciences in China

基　　金：国家科技重大专项(09-Y30F01-9001-20/22);国家民用空间基础设施陆地观测卫星共性应用支撑平台项目(Y930280A2F);中国农业科学院科技创新工程(CAAS-2022-IARRP-GJ2022-20-2)。

摘　　要：利用高光谱遥感数据结合统计建模是当前土壤有机质(SOM)含量高光谱估测的主要方法。为了探讨SOM含量高光谱估测适宜的光谱变换方法和光谱分辨率,以黑龙江省建三江黑土区土壤样本为研究对象,采用SR-6500便携式光谱仪在实验室测量土样的光谱反射率。对土壤光谱数据重采样为1、5、10、20、30、40、50、60、70、80、90、100 nm共12种光谱分辨率,经过Savitzky-Golay光谱曲线平滑处理后,将光谱反射率R进行反射率倒数1/R、对数logR、倒数对数log(1/R)、对数倒数1/logR、一阶导数R′、倒数一阶导数(1/R)′、对数一阶导数(logR)′、倒数对数一阶导数[log(1/R)]′和对数倒数一阶导数(1/logR)′共10种光谱变换;利用多元线性逐步回归(MLSR)和偏最小二乘回归(PLSR)的方法建立SOM含量估测模型。结果表明:(1)1/R和(1/R)′光谱变换对于提高SOM含量估测精度的效果较好,其中1/R光谱变换的SOM含量估测精度R^(2)_(val)均高于0.87,(1/R)′光谱变换的SOM含量估测精度R^(2)_(val)均高于0.90;(2)5、10 nm光谱分辨率对于1/R和(1/R)′光谱变换下的MLSR与PLSR估测SOM含量的精度都较高,为较适宜的光谱分辨率;(3)SOM含量估测的光谱分辨率并非越高越好,适度的降低光谱分辨率以及选择合适的光谱变换方法,不仅可以减少数据处理的工作量,也可以提高SOM含量的估测精度。The use of hyperspectral remote sensing data combined with statistical modeling is currently the main method for hyperspectral estimation of soil organic matter(SOM)content.In order to explore the optimal spectral transformation and spectral resolution for SOM content estimation,the study took soil samples from the black soil area of Heilongjiang province as the research object,and the spectral reflectance of the soil sample were measured by the SR-6500 portable spectrometer in the laboratory.Spectral reflectance was resampled to 12 kinds of spectral resolution such as 1,5,10,20,30,40,50,60,70,80,90 and 100 nm,smoothed by Savitzky-Golay,and subjected to 10 kinds of spectral transformation such as reflectance reciprocal(1/R),logarithm(logR),reciprocal logarithm[log(1/R)],logarithm reciprocal(1/logR),first-order differential(R′),reciprocal first-order derivative[(1/R)′],logarithmic first-order derivative[(logR)′],reciprocal logarithm first-order derivative[log(1/R)]′and logarithmic first-order derivative[(1/logR)′].The SOM content prediction model was established by using multiple linear stepwise regression(MLSR)and partial least square regression(PLSR)methods.The results showed:(1)spectral transformation of 1/R and(1/R)′performed higher and more stable SOM content estimation accuracy with R^(2 )_(val)≥0.87 and R^(2 )_(val)≥0.90,respectively;(2)spectral resolution of 5 and 10 nm had higher estimating accuracy by MLSR and PLSR under 1/R and(1/R)′spectral transformation,which were selected as suitable spectral resolutions;(3)the spectral resolution used for SOM content estimation was not as high as possible.Appropriate reduction of spectral resolution and selection of appropriate spectral transformation can not only reduce the workload of data processing,but also improve the SOM content estimation accuracy.

关 键 词：土壤有机质含量 高光谱 光谱分辨率 光谱变换 估测 

分 类 号：S153.621[农业科学—土壤学]